An efficient rapid system for profiling the cellular activities of molecular libraries

Jonathan S. Melnick*†, Jeff Janes†‡, Sungjoon Kim†‡, Jim Y. Chang‡, Daniel G. Sipes‡, Drew Gunderson‡, Laura Jarnes‡, Jason T. Matzen‡, Michael E. Garcia‡, Tami L. Hood‡, Ronak Beigi‡, Gang Xia‡, Richard A. Harig‡, Hayk Asatryan‡, S. Frank Yan‡, Yingyao Zhou‡, Xiang-Ju Gu‡, Alham Saadat‡, Vicki Zhou‡, Frederick J. King‡, Christopher M. Shaw‡, Andrew I. Su‡, Robert Downs‡, Nathanael S. Gray‡, Peter G. Schultz*§, Markus Warmuth‡§, and Jeremy S. Caldwell‡§

‡Genomics Institute of the Novartis Research Foundation, 10675 John Jay Hopkins Drive, San Diego, CA 92121; and *The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037

Contributed by Peter G. Schultz, December 29, 2005 Rapid quantitative methods for characterizing small molecules, of disorders, most notably cancer (4, 5). For example, spontaneous peptides, proteins, or RNAs in a broad array of cellular assays translocations in which domains become fused to would allow one to discover new biological activities associated other genes (such as Bcr, Tel, and NPM) have been identified as the with these molecules and also provide a more comprehensive etiological basis for multiple B cell lymphomas, including the profile of drug candidates early in the drug development process. Philadelphia chromosome Bcr-Abl and chronic myelogenous leu- Here we describe a robotic system, termed the automated com- kemia (6, 7). It remains a significant challenge to develop selective pound profiler, capable of both propagating a large number of cell inhibitors for tyrosine kinases given their homology and potential lines in parallel and assaying large collections of molecules simul- structural plasticity. To this end, we have established a panel of taneously against a matrix of cellular assays in a highly reproduc- tyrosine-kinase-dependent cellular assays by creating stably ex- ible manner. To illustrate its utility, we have characterized a set of pressed Tel-tyrosine kinase fusions representing each branch of the 1,400 kinase inhibitors in a panel of 35 activated tyrosine-kinase- tyrosine kinase phylogeny. A library of Ͼ1,000 kinase-directed dependent cellular assays in dose–response format in a single heterocycles was then profiled against this panel of cellular assays. experiment. Analysis of the resulting multidimensional dataset

The resulting dataset was examined for global structure–activity CELL BIOLOGY revealed subclusters of both inhibitors and kinases with closely relationships (SARs), potency and selectivity correlations, and correlated activities. The approach also identified activities for the opportunistic side activities. p38 inhibitor BIRB796 and the dual src͞abl inhibitor BMS-354825 and exposed the expected side activities for Glivec͞STI571, includ- Results and Discussion ing cellular inhibition of c- and platelet-derived growth factor Construction of an ACP System. A robotic system was designed that receptor. This methodology provides a powerful tool for unravel- combines cell culture processes with compound screening capabil- ing the cellular biology and molecular pharmacology of both ities to create a massively parallel compound profiling system. This naturally occurring and synthetic chemical diversity. system mechanically integrates all steps involved in cell line main- tenance, including propagation (passaging, splitting, cell density, drug discovery ͉ high-throughput screening ͉ tyrosine kinase and viability determination) with the steps involved in performing compound screens (cell dispensing into microtiter plates, com- he ability to simultaneously interrogate the activities of a library pound addition, incubation, detection, reagent addition, and plate Tof molecules against a large panel of cellular assays would reading͞imaging), as described in Materials and Methods and as provide a rapid efficient means to begin to characterize and depicted in Fig. 1. The design specifications include the ability to correlate the biological properties of both natural and synthetic rapidly profile thousands of compounds in dose–response format in chemical diversity. For example, libraries of noncoding RNAs, miniaturized 1,536-well plate format against hundreds of roboti- mutant growth factors, small molecule kinase inhibitors, or even cally maintained cell-based assays in a highly reproducible way. The existing drugs could be assayed for their potency and selectivity in focus on cell maintenance required strict environmental control of pathway-based or receptor screens or toxicity and metabolic sta- humidity, temperature, sterility, and cell line cross-contamination, bility in diverse cell types to discover a new biological activity or completely distinct from those confronted using biochemical͞ optimize the pharmacological properties of a molecule (1–3). protein assays. The combination of automated tissue culture and Although whole-cell systems represent an attractive milieu to robotic assay technologies enables small-molecule screens to be characterize gene and small-molecule function, no robust and performed on an unprecedented scale in cell-based formats. This systematic method exists to correlate chemical structure and bio- system can also be adapted to screen other molecular libraries, logical activity across a large number of molecules and cellular including secreted peptides and proteins, antibodies, cDNAs, and assays. To address this problem, we have developed an approach siRNAs against collections of cellular assays targeting either spe- that affords rapid cost-effective broad-based cellular profiling in cific gene families (kinases, G-protein-coupled receptors, pro- parallel against molecular libraries. An industrial-scale automated teases, nuclear hormone receptors, etc.), signaling pathways (us- compound profiling (ACP) system has been designed, which con- ing reporter gene, phenotypic, and image-based readouts), or sists of an automated tissue culture system for propagating cell lines, pharmacological properties (metabolic stability, cellular toxicity, or integrated with a system for automatically performing miniaturized transport). cell-based assays in 384- or 1,536-well microplates. The ACP can rapidly test thousands of arrayed molecules, in replicates, in dose– response format against hundreds of unique cellular assays in a Conflict of interest statement: No conflicts declared. single experiment. Freely available online through the PNAS open access option. To demonstrate this capability, we focused on the problem of Abbreviations: ACP, automated compound profiling; SAR, structure–activity relationship. identifying selective small-molecule inhibitors of protein tyrosine †J.S.M., J.J., and S.K. contributed equally to this work. kinases. Tyrosine kinases play a key role in many cellular processes, §To whom correspondence may be addressed. E-mail: [email protected], mwarmuth@ including development, differentiation, and proliferation; misregu- gnf.org, or [email protected]. lation of tyrosine kinase expression and activity leads to a number © 2006 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511292103 PNAS ͉ February 28, 2006 ͉ vol. 103 ͉ no. 9 ͉ 3153–3158 Downloaded by guest on October 1, 2021 cell line model has frequently been used to demonstrate the differential cytotoxicity of kinase inhibitors against transformed IL-3-independent cells versus parental Ba͞F3 cells grown in the presence of IL-3 (16–18). After selection for puromycin and IL-3 withdrawal, 35 of the 81 fusion kinase constructs gave rise to unique cell populations exhibiting growth factor independence. Fig. 2b summarizes the identity of the kinases found to induce growth factor independence in Ba͞F3 cells. All cells were further validated as described (Ma- terials and Methods) by Western blot and RT-PCR͞sequencing of the respective fusion and by using sets of reference compounds for each individual kinase. In the latter case, two independently gen- erated cultures of IL-3-independent cells were required to show similar responses to reference compounds. To determine whether kinase inhibitor activity was accurately reflected by the Tel-kinase- transformed Ba͞F3 cells, a collection of Ϸ80 tyrosine kinase inhibitors with known activities and selectivities was tested in dose–response format against the cell panel. For example, the Sugen c-Met inhibitor PHA-665752 inhibited the growth of Ba͞ F3͞Tel-Met with an IC50 of 80 nM and showed no toxicity toward the rest of the panel, in agreement with published results (19). Likewise, the kinase-insert domain receptor (KDR) inhibitor AAL993 (20) shows selective nanomolar potency against Ba͞F3͞ Fig. 1. The components of the ACP system. (A) Three custom environmen- Tel-KDR and the homologous VEGF receptor family member tally controlled (temperature, CO2, and humidity) 486-plate or flask capacity FLT4. Known BCR-ABL inhibitors, STI571 and AMN107, were incubators. (B) Combination flask to 1,536-well plate cell and reagent dis- also tested against the Ba͞F3 Tel-TK panel and a panel of recom- penser. (C) Control station consisting of a computer running custom schedul- binant . As shown in Fig. 2c, both STI571 and AMN107 ing software. (D) Perkin–Elmer ViewLux plate reader. (E) Olympus (Melville, inhibited Bcr-Abl, c-Kit, and PDGF-RA͞B with IC50s between 1 NY) inverted microscope with an automated stage and Molecular Devices and 100 nM as expected [data are shown as percent activity of the METAMORPH software. (F) Compound transfer station (typically 5- to 50-nl trans- versus control at a single concentration of 10 ␮M and was volumes). (G) Proprietary tissue culture (TC) station. (H) BD Biosciences compared to the inhibitory concentrations (IC50) generated by FACSArray Bioanalyzer with custom sampling tray. (I) Staubli RX130 anthro- ͞ pomorphic robot with a custom end effector (gripper). using Ba F3 cell proliferation assays]. In addition, AMN107 also inhibited the proliferation of Ba͞F3 cells expressing Tel fusions of the receptor kinases EphB1, B2, and B4, suggesting that the Generation and Validation of a Tel-Fused Tyrosine Kinase Panel in increased potency of AMN107 compared with STI571 results in Ba͞F3 Cells. To illustrate the utility of ACP, we chose to profile a somewhat decreased selectivity. Interestingly, several discrepancies panel of kinase-dependent cellular assays against a collection of between the cellular and enzymatic assays were found. For exam- kinase-directed heterocycles. Although parallel interrogations of ple, although very potent on both PDGF-RA͞B in cellular assays, the kinome against chemical space have been performed by using both STI571 and AMN107 only partially inhibited PDGF-RA͞Bin in vitro biochemical and phage-display assays, the cell-based format the enzymatic assays. A similar observation was made for EphB4 described here assays the physiological conformation of the kinase and AMN107. This is best explained by the fact that both inhibitors in the presence of other cellular components, cell permeability, and were shown to bind the inactive conformations of kinases, which nonselective cellular toxicity (6, 7). To profile tyrosine kinases in a might constitute only a fraction of the enzyme preparations used. cell-based format, we made use of the well established fact that Finally, some of the activities seen in the enzymatic assays could not kinases can be constitutively activated by genomic rearrangements be reproduced using cellular assays. IC50 measurements for these leading to the juxtaposition of a fusion partner to a kinase (8). enzymes would be required to draw more detailed conclusions. Several chimeric tyrosine kinases, including Bcr-Abl, NPM-Alk, and ETV6-NTRK3, have been described and shown to be causative Kinase Profiling Experiment. A chemical library totaling 1,400 unique small molecules targeting tyrosine kinases was selected that to human cancer and hematopoietic malignancies. Frequently includes purines, pyrimidines, benzoimidazoles, and quinazolines found fusion partners include structural proteins as well as tran- (21). The library was supplemented with an additional 10 known scription factors or genes of unknown functions. The most frequent ͞ tyrosine kinase inhibitors as controls and plated into 1,536-well fusion partner for tyrosine kinases is ETV6 Tel, a gene that has microtiter plates prediluted into dose–response format ranging been found in chimeras with both cytosolic and receptor tyrosine from 10 ␮M down to 3 nM. Each individual population of Ba͞ kinases (Abl, NTRK3, PDGFR, and Jak2) (9–12), suggesting that ͞ F3-TK cells was seeded on the automated tissue culture station, fusion to ETV6 Tel might be a generally applicable strategy to propagated for two to three passages; checked for cell viability, activate tyrosine kinases. growth rates, and cell density; and then plated into 1,536-well assay ͞ To generate a cDNA library of kinases fused to ETV6 Tel, a plates on the robotic platform. The preplated compounds were retroviral plasmid based on a pMSCV backbone was constructed transferred to assay plates in register (1:1) by using a 1,536-pin that allows rapid in-frame cloning of a kinase domain upstream of transfer device, incubated overnight, then assayed for cell viability a Tel cassette and downstream of a Myc tag using a Gateway as described in Materials and Methods. The experiment was re- cassette system (Fig. 2a). A total of 81 Tel fusion vectors repre- peated in triplicate resulting in Ϸ1.5 million data points. To senting individual kinases, as well as the control non-Tel fusion compare the data quality generated on the robotic system with data Bcr-Abl, were established and used to transform Ba͞F3 cells as generated manually with workstations, a subset (936) of the com- described in Material and Methods.Ba͞F3 is a hematopoietic cell pounds was tested against 20 of the Ba͞F3-TK cells in 384-well assay line that depends on IL-3 for proliferation and survival but is plate format in duplicate. Analysis of the resulting datasets showed rendered IL-3-independent by transformation with known tyrosine that for both the ACP and the manual method, Ͼ95% of the data kinase oncogenes, e.g., Bcr-Abl, Flt3, and NPM-Alk (13–15). This had coefficients of variation of Ͻ10%. By defining outliers as those

3154 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511292103 Melnick et al. Downloaded by guest on October 1, 2021 CELL BIOLOGY

Fig. 2. Ba͞F3 construction. (a) Retroviral construct for Ba͞F3-TK library. pTK is based on the pMSCV (Clontech) backbone. The first 1 kb of ETV6, with an artificial myristoylation sequence (mTel), is located upstream of the Gateway reading frame B cassette (Invitrogen), which is followed by the Myc tag sequence EQKISEEDL and a stop codon. This is linked by the HCV IRES to puromycin in which the stop codon has been replaced by the luciferase coding sequence from pLucN2 (Clontech). The expressed protein, a functional dimer, is detailed to the left. Dimerization of the mTel domains bring the two fused kinase domains into proximity, leading to autophosphorylation and activation. (b) The tyrosine kinome. The tyrosine kinase-dependent Ba͞F3 lines used in this screen are highlighted in green boxes against a phylogeny of all 90 tyrosine kinases reproduced courtesy of Cell Signaling Technology (Danvers, MA). (c) Cellular and biochemical assays. The Bcr-Abl-targeted kinase inhibitors Glivec and AMN107 were tested against the kinase cell line panel (Ba͞F3) and the purified enzyme (in vitro). Data for Ba͞F3 are listed as the IC50 in ␮M, whereas the in vitro data are listed as the percent enzymatic activity remaining at 10 ␮M compound. The boxes are colored in green (potent inhibition), black (mild inhibition), or red (little to no inhibition).

IC50s with Ͼ3-fold or less than one-third of the average IC50, only displayed a 50% growth inhibition (GI50) Ͻ 10 ␮M, giving a two outliers of 936 data points were observed in the ACP dataset ‘‘non-specificity count.’’ Every test point for which the GI50 was Ͻ10 relative to the manual dataset. ␮M was plotted with the negative log of the GI50 (pGI50)onthe ordinate and the non-specificity count of that compound on the Global Profile Data Analysis. An examination of the 1,400 com- abscissa. Although the global dataset of 935 nontoxic compounds pounds tested demonstrated that only 30 (2.1%) were toxic to was uninformative, inspection of clusters of structurally related wild-type IL-3-dependent Ba͞F3 cells Ͻ1 ␮M; Ϸ400 compounds compounds revealed 9 of 14 classes that showed a modest corre- showed slight of toxicity at micromolar concentrations. Interest- lation between increases in potency and selectivity (Fig. 3a). ingly, 282 (20.1%) compounds did not affect the activity of any Next, we asked whether chemical similarity was a predictor of TK-dependent cell line Ͻ5 ␮M; however, every kinase on the panel biological activity within this dataset. For each pair of compounds, was inhibited by at least one compound. Finally, 83 (5.9%) com- the chemical similarity was computed by the Tanimoto similarity of pounds selectively inhibited a single kinase in the panel. 512-bit Daylight fingerprints. The similarity in biological response Typically, as the potency of a compound increases, parallel gains between two compounds was calculated as the ordinary Pearson in selectivity occur (22, 23). Regression analysis was used to correlation of the vectors composed of the pGI50 values across the determine whether the profiling data are consistent with this 36 assays. The plot of these two metrics displays a very strong premise. Compounds were classified according to specificity by relationship between the similarity in chemical structure and sim- counting the number of the 36 assays in which each compound ilarity in biological activity (Fig. 3b). Although the relationship

Melnick et al. PNAS ͉ February 28, 2006 ͉ vol. 103 ͉ no. 9 ͉ 3155 Downloaded by guest on October 1, 2021 pounds in the pair, usually because the compound has little or no activity in all of the kinase assays. Such low variance causes correlation-based distance measures to be brittle, responding dra- matically to slight changes in the measured GI50 for a single assay. Another group of outliers are compound pairs in which a small structural change leads to a slight general cytotoxicity. Because this cytotoxicity is reflected in the GI50 for all 36 kinase assays, the cumulative effect is to produce large differences in biological profile. Finally, there are a smaller number of outliers that appear to be genuine exceptions to the SAR hypothesis, in which small changes of chemical structure lead to large changes in biological profile. These three categories are comingled in Fig. 3b, and inspection of the individual profiles (not shown, but analogous to those in Figs. 2c and 4b) is necessary to distinguish them. An SAR dendrogram was created to relate kinase similarity as a function of compound activity. Distances between kinase pairs in ‘‘profile space’’ were calculated as the Euclidean distance between the vectors composed of the pGI50 values of the 935 nontoxic compounds, with inactive compounds being assigned a surrogate GI50 of 10 ␮M. This distance matrix was subjected to agglomerative clustering by using the complete linkage method. The data are represented as a dendrogram (Fig. 5, which is published as sup- porting information on the PNAS web site). As expected, highly homologous kinases are most often inhibited comparably by small molecules. The most notable exception is the close proximity of Flt3 to the Trk kinases in the SAR dendrogram versus the sequence- based tree. This is likely because these kinases share a critical ‘‘gatekeeper’’ phenylalanine side chain in the ATP-binding center (see Fig. 2b). In general, the SAR dendrograms can be used as tools to guide knowledge-based target selection and analyze multipa- rameter datasets obtained from compound profiling. Targets in close proximity on the dendrogram are more likely to exhibit comparable inhibition by a particular group of inhibitors than targets that are more distant. The dendrogram can also be used to select and pursue targets that are inhibited by similar chemotypes.

Previously Uncharacterized Activities. The data were next scanned to identify whether any known drugs and well characterized com- pounds from the chemical collection displayed any previously unknown activities. The data were configured into a heat map by hierarchical clustering for ease of analysis (Fig. 4a). The data were examined to find targets for existing molecules and to understand the activity of molecules on related targets. For Glivec, activity against Bcr-Abl, platelet-derived growth factor, and c-kit were all Fig. 3. Global data analysis. (a) Selectivity͞potency correlation. The com- manifest (see Fig. 4b). Interesting cross-activities for other less well pounds were clustered based on chemical structure (Pipeline Pilot, Scitegic, characterized kinase inhibitors currently under clinical investigation San Diego) to yield 94 clusters. The test points for 14 clusters (those with 20 or were also discovered. For example, BIRB796, a kinase inhibitor for more compounds) were plotted broken out by cluster to assess the specificity– all p38 kinase isoforms (24, 25), was also shown to inhibit Tie2 and potency relationships for focused structural classes. The clusters are numbered from 1 to 14, where blue dots represent compounds, and are distributed by to a lesser extent Tie1 (Fig. 4b). These activities were confirmed to specificity (abscissa) and potency (GI50; ordinate). (b) Relationship of struc- be target-related and not dependent on inhibition of p38 by tural similarity and biological similarity. Each marker represents a pair of Western blot of both Tel-Tie1 and Tel-Tie2 immunoprecipitated compounds, with the pairwise chemical distance (Tanimoto coefficient) on the from Ba͞F3 cells and probed with antiphosphotyrosine antibodies abscissa and the pairwise biological profile distance (Pearson correlation (data not shown). Tie2 has recently been implicated in coefficient) on the ordinate. The line shown is the ordinary least-squares fit. as well as stem cell quiescence and mobilization during chemother- The triangle encloses the outliers, which are discussed in the text. Markers are apy (24–27). Therefore, expansion to Tie2 as a target could colored according to an alternative measure of biological profile similarity, potentially increase the utility of this compound. the cosine (or ‘‘uncentered’’) correlation coefficient. Although BIRB796 displayed high selectivity versus other ty- rosine kinases in our experiment, BMS-354825, a dual inhibitor of Src and Abl (28), was shown to inhibit multiple other kinases, between chemical and biological similarity is strong, it is clearly including several Ephrin receptors (B1, B2, and B4). Ephrin nonlinear and noisy. One source of this nonlinearity is the folded receptors have been implicated in both tumor angiogenesis and nature of Daylight fingerprint bitmaps, which causes the similarity growth and survival of tumor cells (29–31). BMS-354825 is cur- for unrelated compounds to cluster around a Tanimoto coefficient rently being tested in humans for its potential to overcome Glivec- of 0.5. The most interesting outliers are those that have a high resistant Bcr-Abl-positive chronic mylogenous and acute lympho- chemical but a low biological similarity (enclosed by the triangle in blastic leukemia (32). Our results indicate the potential for Fig. 3b). An inspection of the biological profiles of these outliers expansion into other tumor types. In summary, these data demon- reveals three general classifications. There are outliers in which the strate the power of the aforementioned ACP in identifying novel biological profile vector has low variance for one or both com- indications for known drugs and drug candidates, a fact that will

3156 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511292103 Melnick et al. Downloaded by guest on October 1, 2021 CELL BIOLOGY

Fig. 4. New targets. (a) Heat map of ACP data. IC50 values from the triplicate screens were averaged and clustered by using The Institute of Genome Research Multiple Array Viewer according to the average linking method of the Euclidian distance hierarchical clustering metric to generate the heat map showninthe lower left. The scale ranges from 0 (green) to 10 ␮M (red), as shown in the right-hand color triangle. The dendrogram associated with the heat map, showing kinase linkage, is enlarged in b, and selected inhibitors referred to in the text are expanded underneath. (b) Activities. Selected compounds from the ACP profiling experiment suggest alternate targets for several compounds. Here, the pan-p38 kinase inhibitor BIRB796 also inhibits Tie1 and Tie2. The dual src- inhibitor BMS-354825 also inhibits the Ephrins. Glivec (STI571) inhibits BCR-ABL, c-kit, and PDGF-RA͞B.

become increasingly important given that market sizes for molec- biological space. Automated kinase profiling expands SAR from ularly targeted therapies based on genetic lesions might generally be one target against a set of compounds to many targets against many rather small. compounds, thus providing a more comprehensive dataset. In the case of the kinome, this type of information facilitates the process Conclusion of ‘‘kinase hopping,’’ to determine which scaffolds are most likely Herein, we have profiled a collection of 1,400 small-molecule kinase to have activity on new targets of interest. inhibitors in a dose–response format against an array of 35 cell- Opportunistic cellular profiling is the preclinical corollary to based tyrosine kinase assays in a single experiment using a highly serendipitous clinical profiling, which led to discoveries such as efficient profiling technology. The ACP provides a mechanism for Viagra’s use in erectile dysfunction (despite its originally intended systematic 2D combinatorial screening of chemical space against use in heart disease) or the demonstration that cholesterol synthesis

Melnick et al. PNAS ͉ February 28, 2006 ͉ vol. 103 ͉ no. 9 ͉ 3157 Downloaded by guest on October 1, 2021 inhibitors, statins, reduce CD69 T cell antigen levels in T cells, which daughter flasks for pooling (if the number of cells required is may extend the statin’s benefits to immune regulation (33–35). As unavailable, the cell propagation protocol is enacted). When daugh- demonstrated here, the ACP experiment identified PDGFR and ter flasks have incubated for a defined amount of time, they are c-kit as side activities for Glivec, which are under investigation for pooled in a matrix fashion into several empty recipient flasks. An alternative treatments of asthma and gastrointestinal stromal dis- equal volume from each daughter cell flask will be deposited into order (GIST). Similarly, the activities identified for the p38 kinase each recipient flask. A sample from one pooled flask will be inhibitor BIRB796 and dual src͞abl inhibitor BMS-354825 may counted (only one sample is necessary, because all of the flasks have prove useful as tools to validate Tie2 and the Ephrins as drug targets been adjusted to the same cell density by the pooling process). All in angiogenesis. of the flasks will have their volume adjusted to the desired cell ACP profiling of molecular libraries against diverse cellular density for plating. This pooling function, although complex, obvi- assays can be applied to many other problems as well. For example, ates the requirement for a common pooling container, thus de- it may be possible to identify novel ligands for whole panels of creasing the possibility of contamination. The flasks are then moved to the cell dispenser, where cells are pulled from flasks and orphan G-protein-coupled receptors by profiling collections of ␮ ͞ diverse lipid, metabolite, and neuropeptide hormone libraries. It dispensed into 1,536-well assay plates, typically 5 l well. The plates may also be possible to identify combinations of drugs that act of cells are placed into an incubator and the empty flasks discarded. synergistically against panels of patient-derived tumor cell lines. For For the assay described in this paper, compounds are introduced to the system in 1,536-well plates through the room-temperature pharmacogenomics, disease-associated SNPs identified by haplo- hotels. Cell plates are requested at 105 cells͞ml, 5 ␮l͞well, in type mapping can be engineered into SNP-dependent cellular 1,536-well plates, for each of the 35 Ba͞F3 cell lines. The cell assays and profiled against panels of preclinical drug candidates to requests are processed as described above. After cell plating, prospectively match patient variants with treatment. The configu- preplated compounds in single-dose or dose–response format are ration of the ACP also allows screens for ligands with enhanced transferred from the compound plates to the assay plates by using potency, selectivity, stability, or expression levels from evolved the 1,536 pintool. The assay plates are incubated in the environ- protein libraries. mentally controlled (37°C, 95% humidity, 5% CO2) incubators for 48 h (a custom lid design minimizes edge effects from evaporation Materials and Methods and enables 5-␮l cell-based assays to be incubated for up to 5 days). Automated Cellular Assays. The assay portion of the ACP consists of After incubation, 5 ␮l͞well Cell Titer Glo (Promega) is added with the same control software as the automated tissue culture station, the reagent dispenser, incubated for 10 min at room temperature, a noncontact 1,536-well reagent dispenser (dispense range of 250 nl and luminescence quantitated with the Perkin–Elmer ViewLux. A to 5 ␮l at 5% coefficient of variation (CV), dispense time of Ͻ1 luminescence value less than the control value indicates a decrease min͞plate at 5 ␮l), a 1,536-pin transfer device (validated at 9 nl and in either the number of cells or cell viability. Ͼ10% CVs), room-temperature hotels, the same incubators as described above, and a Perkin–Elmer ViewLux plate reader. To Supporting Text. Protocols detailing the automatic propagation of request cells for an assay, the operator inputs the desired final cell lines using the ACP, as well as the construction of the number of assay plates, cell density, and well volume. The auto- Ba͞F3͞Tel-kinase library, are presented in Supporting Text, which mated tissue culture station calculates the correct number of is published as supporting information on the PNAS web site.

1. Plavec, I., Sirenko, O., Privat, S., Wang, Y., Dajee, M., Melrose, J., Nakao, B., 19. Christensen, J. G., Schreck, R., Burrows, J., Kuruganti, P., Chan, E., Le, P., Hytopoulos, E., Berg, E. L. & Butcher, E. C. (2004) Proc. Natl. Acad. Sci. USA Chen, J., Wang, X., Ruslim, L., Blake, R., et al. (2003) Cancer Res. 63, 101, 1223–1228. 7345–7355. 2. Rabow, A. A., Shoemaker, R. H., Sausville, E. A. & Covell, D. G. (2002) J. Med. 20. Manley, P. W., Bold, G., Bruggen, J., Fendrich, G., Furet, P., Mestan, J., Chem. 45, 818–840. Schnell, C., Stolz, B., Meyer, T., Meyhack, B., et al. (2004) Biochim. Biophys. 3. Butcher, E. C., Berg, E. L. & Kunkel, E. J. (2004) Nat. Biotechnol. 22, Acta 1697, 17–27. 1253–1259. 21. Ding, S., Gray, N. S., Ding, Q. & Schultz, P. G. (2001) J. Org. Chem. 66, 4. Blume-Jensen, P. & Hunter, T. (2001) Nature 411, 355–365. 8273–8276. 5. Hunter, T. (1998) Harvey Lect. 94, 81–119. 22. Gallion, S. L. & Qian, D. (2005) Curr. Opin. Drug Discov. Dev. 8, 638–645. 6. Seidler, J., McGovern, S. L., Doman, T. N. & Shoichet, B. K. (2003) J. Med. 23. Gill, A. L., Frederickson, M., Cleasby, A., Woodhead, S. J., Carr, M. G., Chem. 46, 4477–4486. Woodhead, A. J., Walker, M. T., Congreve, M. S., Devine, L. A., Tisi, D., et 7. Knight, Z. A. & Shokat, K. M. (2005) Chem. Biol. 12, 621–637. al. (2005) J. Med. Chem. 48, 414–426. 8. Krause, D. S. & Van Etten, R. A. (2005) N. Engl. J. Med. 353, 172–187. 24. Regan, J., Breitfelder, S., Cirillo, P., Gilmore, T., Graham, A. G., Hickey, E., 9. Golub, T. R., Barker, G. F., Lovett, M. & Gilliland, D. G. (1994) Cell 77, Klaus, B., Madwed, J., Moriak, M., Moss, N., et al. (2002) J. Med. Chem. 45, 307–316. 2994–3008. 10. Papadopoulos, P., Ridge, S. A., Boucher, C. A., Stocking, C. & Wiedemann, 25. Pargellis, C., Tong, L., Churchill, L., Cirillo, P. F., Gilmore, T., Graham, A. G., L. M. (1995) Cancer Res. 55, 34–38. Grob, P. M., Hickey, E. R., Moss, N., Pav, S. & Regan, J. (2002) Nat. Struct. 11. Lacronique, V., Boureux, A., Valle, V. D., Poirel, H., Quang, C. T., Biol. 9, 268–272. Mauchauffe, M., Berthou, C., Lessard, M., Berger, R., Ghysdael, J., et al. (1997) 26. Oliner, J., Min, H., Leal, J., Yu, D., Rao, S., You, E., Tang, X., Kim, H., Meyer, Science 278, 1309–1312. S., Han, S. J., et al. (2004) Cancer Cell 6, 507–516. 12. Eguchi, M., Eguchi-Ishimae, M., Tojo, A., Morishita, K., Suzuki, K., Sato, Y., 27. Kobayashi, H. & Lin, P. C. (2005) Front. Biosci. 10, 666–674. Kudoh, S., Tanaka, K., Setoyama, M., Nagamura, F., et al. (1999) Blood 93, 28. Lombardo, L. J., Lee, F. Y., Chen, P., Norris, D., Barrish, J. C., Behnia, K., 1355–1363. Castaneda, S., Cornelius, L. A., Das, J., Doweyko, A. M., et al. (2004) J. Med. 13. Daley, G. Q. & Baltimore, D. (1988) Proc. Natl. Acad. Sci. USA 85, 9312–9316. Chem. 47, 6658–6661. 14. Bai, R. Y., Dieter, P., Peschel, C., Morris, S. W. & Duyster, J. (1998) Mol. Cell. 29. Kullander, K. & Klein, R. (2002) Nat. Rev. Mol. Cell Biol. 3, 475–486. Biol. 18, 6951–6961. 30. Brantley-Sieders, D. M. & Chen, J. (2004) Angiogenesis 7, 17–28. 15. Hayakawa, F., Towatari, M., Kiyoi, H., Tanimoto, M., Kitamura, T., Saito, H. 31. Brantley-Sieders, D., Parker, M. & Chen, J. (2004) Curr. Pharm. Des. 10, & Naoe, T. (2000) Oncogene 19, 624–631. 3431–3442. 16. Weisberg, E., Boulton, C., Kelly, L. M., Manley, P., Fabbro, D., Meyer, T., 32. Shah, N. P. (2005) Hematology 1, 183–187. Gilliland, D. G. & Griffin, J. D. (2002) Cancer Cell 1, 433–443. 33. Kunkel, E. J., Plavec, I., Nguyen, D., Melrose, J., Rosler, E. S., Kao, L. T., 17. Shah, N. P., Tran, C., Lee, F. Y., Chen, P., Norris, D. & Sawyers, C. L. (2004) Wang, Y., Hytopoulos, E., Bishop, A. C., Bateman, R., et al. (2004) Assay Drug Science 305, 399–401. Dev. Technol. 2, 431–441. 18. Weisberg, E., Manley, P. W., Breitenstein, W., Bruggen, J., Cowan-Jacob, 34. Boolell, M., Allen, M. J., Ballard, S. A., Gepi-Attee, S, Muirhead, G. J., Naylor, S. W., Ray, A., Huntly, B., Fabbro, D., Fendrich, G., Hall-Meyers, E., et al. A. M., Osterloh, I. H. & Gingell, C. (1996) Int. J. Impot. Res. 8, 47–52. (2005) Cancer Cell 7, 129–141. 35. Padma-Nathan, H. (1999) Am J. Cardiol. 84, 18N–23N.

3158 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0511292103 Melnick et al. Downloaded by guest on October 1, 2021